CombiMOTS: Combinatorial Multi-Objective Tree Search for Dual-Target Molecule Generation

Thibaud Southiratn, Bonil Koo, Yijingxiu Lu, Sun Kim

TL;DR: CombiMOTS is a framework that designs dual-target compounds using industrial building blocks and Pareto MCTS for favorable multi-objective trade-offs and improved synthesizability.

Abstract:

Dual-target molecule generation, which focuses on discovering compounds capable of interacting with two target proteins, has garnered significant attention due to its potential for improving therapeutic efficiency, safety and resistance mitigation.
Existing approaches face two critical challenges.
First, by simplifying the complex dual-target optimization problem to scalarized combinations of individual objectives, they fail to capture important trade-offs between target engagement and molecular properties.
Second, they typically do not integrate synthetic planning into the generative process.
This highlights a need for more appropriate objective function design and synthesis-aware methodologies tailored to the dual-target molecule generation task.
In this work, we propose CombiMOTS, a Pareto Monte Carlo Tree Search (PMCTS) framework that generates dual-target molecules.
CombiMOTS is designed to explore a synthesizable fragment space while employing vectorized optimization constraints to encapsulate target affinity and physicochemical properties.
Extensive experiments on real-world databases demonstrate that CombiMOTS produces novel dual-target molecules with high docking scores, enhanced diversity, and balanced pharmacological characteristics, showcasing its potential as a powerful tool for dual-target drug discovery.
The code and data is accessible through \url{https://github.com/Tibogoss/CombiMOTS}.

Lay Summary:

Most traditional drugs follow the "one drug, one target" paradigm, but complex diseases are often related to several targets. As a response, emerging trends aim to design "dual-target drugs". However, existing multi-objective methods oversimplify the task by (i) aggregating objectives into a scalar value, (ii) neglecting real post-hoc feasibility through known synthetic procedures.

To address this, we propose CombiMOTS, a combinatorial framework that (i) reduces a chemical search space to interesting building blocks specific to target protein pairs, and (ii) uses a Pareto Monte Carlo Tree Search algorithm to assemble them into potential dual-inhibitors. Our method effectively finds optimal candidates across flexible objectives, while simultaneously accounting for structural and physicochemical constraints.

Empirical results show strong and favorable trade-offs across multiple target pairs, showcasing both the quantitative and qualitative performance of CombiMOTS. Additional theoretical analysis and broader preliminarily applications are provided, making CombiMOTS a valuable tool for practitioners interested in multi-objective drug design.

Application-Driven Machine Learning: This submission is on Application-Driven Machine Learning.

Link To Code: https://github.com/Tibogoss/CombiMOTS

Primary Area: Applications->Health / Medicine

Keywords: Dual-target Molecule Generation, Fragment-based Drug Discovery, Monte-Carlo Tree Search, Pareto Optimization, Search Space Reduction

Submission Number: 16227